The World Health Organization brought ministers and senior officials from 37 countries to Lisbon on 15-16 July 2026 to develop a shared approach to AI governance in healthcare, after its own assessment revealed a dangerous gap between rapid adoption and the near-total absence of regulatory safeguards. Nearly two-thirds of countries in the European Region already use AI for diagnostics, yet only 8% have a health-specific AI strategy and only 8% have liability standards when those systems fail.
The summit, co-hosted by WHO/Europe and the Portuguese government, gathered representatives from all six WHO regions alongside the European Commission, the World Bank, the Wellcome Trust, the Aga Khan University and the Gates Foundation. Technical sessions on 13 and 14 July brought together researchers, clinicians and international organisations, while the ministerial meetings on 15 and 16 July translated those discussions into a practical agenda for cooperation.
Three pillars for international cooperation
WHO structured the meeting around three pillars: regulation and accountability, tools and infrastructure for safe deployment, and the people and institutions responsible for implementing AI in clinical settings. Rather than promoting a single model, officials explored how countries at different levels of digital development can adapt common governance principles to their own health systems. Discussions covered legal accountability, ethics, data governance, interoperability, workforce preparation and responsible investment.
Participants stressed that effective AI governance will be essential to ensure AI narrows rather than widens inequalities between well-resourced and under-resourced health systems. A separate session involving Angola, Brazil, Cabo Verde, Portugal, São Tomé and Príncipe, and Timor-Leste laid the foundations for cooperation on AI and health among Portuguese-speaking countries.
The governance gap in numbers
WHO's readiness assessment across the European Region exposed the scale of the mismatch. While AI adoption in diagnostics is widespread, dedicated strategies and liability frameworks are almost nonexistent. Only 8% of countries have a health-specific AI for Healthcare strategy, and the same tiny fraction have standards that define accountability when an AI system causes harm. The findings mean that in most countries, patients are already being diagnosed with the help of AI tools that operate without clear rules on who is responsible if something goes wrong.
The urgency of patient-facing AI
WHO Regional Director for Europe, Hans Kluge, said governments need a shared approach to governance, workforce training and equitable access. He warned that "people are increasingly consulting generative AI chatbots about medical symptoms before speaking with healthcare professionals," highlighting the urgency of effective oversight. The comment reflects a growing reality: AI is not only inside clinical workflows but also in patients' hands, often without any quality control or safety net.
Why this matters for healthcare professionals
The Lisbon summit signals that AI governance in healthcare is moving from a policy abstraction to a practical necessity. For clinicians, hospital administrators and health system leaders, the absence of clear liability frameworks creates legal and ethical risk every time an AI-assisted diagnosis is made. The push for common international principles also means that workforce training and interoperability standards are likely to become central to future procurement and accreditation decisions. Healthcare professionals who understand these governance trends will be better positioned to advocate for safe, equitable AI deployment in their own institutions.
Your membership also unlocks:
